Quantum Dynamics of Dissociative Chemisorption of H2 on the

6 hours ago - Resources and Information. Journals A-Z · Books and Reference · Advertising Media Kit · Institutional Sales · ACS Publishing Center · Pr...
1 downloads 0 Views 2MB Size
Subscriber access provided by Nottingham Trent University

C: Surfaces, Interfaces, Porous Materials, and Catalysis

Quantum Dynamics of Dissociative Chemisorption of H on the Stepped Cu(211) Surface 2

Egidius W.F. Smeets, Gernot Füchsel, and Geert-Jan Kroes J. Phys. Chem. C, Just Accepted Manuscript • DOI: 10.1021/acs.jpcc.9b06539 • Publication Date (Web): 23 Aug 2019 Downloaded from pubs.acs.org on August 23, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Quantum Dynamics of Dissociative Chemisorption of H2 on the Stepped Cu(211) Surface Egidius W.F. Smeets,† Gernot F¨uchsel,‡ and Geert-Jan Kroes∗,† †Leiden Institute of Chemistry, Gorlaeus Laboratories, Leiden University, P.O. Box 9502, 2300 RA, The Netherlands ‡Institut f¨ ur Chemie und Biochemie - Physikalische und Theoretische Chemie, Freie Universit¨at Berlin, Takustraße 3, 14195 Berlin, Germany E-mail: [email protected]

1

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Abstract Reaction on stepped surfaces are relevant to heterogeneous catalysis, in which reaction often takes place at the edges of nanoparticles where the edges resemble steps on single crystal stepped surfaces. Previous results on H2 + Cu(211) show that in this system steps do not enhance the reactivity, and raised the question of whether this effect could be in anyway related to the neglect of quantum dynamical effects in the theory. To investigate this we present full quantum dynamical molecular beam simulations of sticking of H2 on Cu(211) in which all important rovibrational states populated in a molecular beam experiment are taken into account. We find that the reaction of H2 with Cu(211) is very well described with quasi-classical dynamics when simulating molecular beam sticking experiments, in which averaging takes place over a large number of rovibrational states and over translational energy distributions. Our results show that the stepped Cu(211) surface is distinct from its component Cu(111) terraces and Cu(100) steps and cannot be described as a combination of its component parts with respect to the reaction dynamics when considering the orientational dependence. Specifically, we present evidence that at translational energies close to the reaction threshold vibrationally excited molecules show a negative rotational quadrupole alignment parameter on Cu(211), which is not found on Cu(111) and Cu(100). The effect arises because these molecules react with a site specific reaction mechanism at the step, i.e., inelastic rotational enhancement, which is only effective for molecules with a small absolute value of the magnetic rotation quantum number. From a comparison to recent associative desorption experiments as well as Born-Oppenheimer molecular dynamics (BOMD) calculations it follows that the effects of surface atom motion and electron hole-pair (ehp) excitation on the reactivity fall within chemical accuracy, i.e., modeling these effect shifts extracted reaction probability curves by less then 1 kcal/mol translational energy. We found no evidence in our fully-state-resolved calculations for the ’slow’ reaction channel that was recently reported for associative desorption of H2 from Cu(111) and Cu(211), but our results for the fast channel are in good agreement with the experiments of H2 + Cu(211).

2

ACS Paragon Plus Environment

Page 2 of 51

Page 3 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

1. Introduction The rate limiting step in heterogeneous catalysis is often a dissociative chemisorption reaction 1,2 . Hydrogen (H2 ) dissociation is important to heterogeneously catalyzed production of syngas and ammonia 3 and has recently gained industrial importance with the production of methanol from CO2 over a Cu/ZnO/Al2 O3 catalysts, in which the rate limiting step is considered to be the dissociation of H2 4–6 . Stepped, kinked or otherwise defective surfaces more closely resemble real catalytic surfaces, as catalyzed reactions tend to proceed at the corners or edges of nanoparticles 7,8 . A better theoretical understanding of the reaction dynamics of H2 dissociation on stepped surfaces could well be a first step to the design of new catalysts from first principles 9 .

H2 reacting on copper surfaces is a prototypical example of a highly activated late barrier system 10–13 . For the flat Cu(111), Cu(110) and Cu(100) surfaces a plethora of experimental 13–24 as well as theoretical 12,25–43 results have been reported that are generally in good agreement with each other. This large body of work has allowed for the development of a chemically accurate description of molecular beam experiments using the semi-empirical specific reaction parameter approach to density functional theory (SRP-DFT) 35 . Recently molecular beam adsorption experiments 44 and associative desorption experiments 45 for H2 reacting on Cu(211) have been reported, allowing for a more stringent comparison between theory and experiment for this system that more closely resembles a catalytic particle. Theoretical reaction dynamics of H2 reacting on stepped or defective surfaces have only been reported sparingly, most notably for D2 on Cu(211) 46 , H2 on Pt(211) 47–51 and H2 on defective Pd(111) 52 .

In our previous work we and others have shown that the Cu(211) surface is less reactive then the Cu(111) surface 46 , which indicates that predictions based on the d-band model of Nørskov and Hammer 53,54 are not always reliable. In the d-band model increased reactiv3

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ity at steps, defects, or otherwise less coordinated surface atoms, is ascribed to a reduced width of the d-band 53,55 and a shift of the center of the d-band towards the Fermi level at these sites. In the case of Cu(211) the breakdown of the d-band model is due to the geometric effect of the lowest barrier to reaction for H2 on Cu(111) not being situated at a top site 46 .

Due to the corrugated nature of the molecule surface interaction and the denser distribution of barriers to reaction it is unclear whether quantum effects can have a significant effect on the reaction dynamics of H2 reacting on Cu(211). Our main goal is to investigate if including quantum effects during the dynamics significantly affects observables such as the macroscopic molecular reaction probability and rotational quadrupole alignment parameters. To this end we will mainly focus on a comparison of fully state-resolved quantum dynamical (QD) and quasi-classical trajectory (QCT) reaction probabilities for H2 incident on Cu(211), and the effect of Boltzmann averaging over all rovibrational states populated in a molecular beam experiment. Employing the time-dependent wave packet (TDWP) method 56,57 , we have carried out QD calculations mainly for H2 . Due to the low mass of H2 quantum effects are presumed to be most prevalent for H2 and energy transfer to the surface during collision is expected to be small. Performing this large body of calculations for D2 would have been much more expensive because its larger mass necessitates the use of denser numerical grids and longer propagation times.

Another aim will be to investigate if the reaction dynamics of H2 dissociation on the stepped Cu(211) differs from the reaction dynamics at low Miller index copper surfaces, for which the reaction dynamics is reasonably similar 30–32,43 . This is relevant because the Cu(211) surface has Cu(111) terraces and Cu(100) steps, and considering this question might thus provide more insight in how a stepped surface can alter reaction mechanisms. Rotational quadrupole alignment parameters for vibrationally excited molecules are similar in behavior for Cu(111) 16,32 and Cu(100) 30,31 . We will investigate whether the same holds for H2 +

4

ACS Paragon Plus Environment

Page 4 of 51

Page 5 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Cu(211).

Recent associative desorption experiments on Cu(111) and Cu(211) 45 , which were in good agreement with earlier theoretical and experimental work 16,30,34,35,46,58 , have shown a never before reported ’slow’ reaction channel to be active for both Cu(111) and Cu(211). In this channel reaction could be facilitated by trapping on the surface and distortion of the surface due to thermal motion forming a reactive site 45 . Our calculations on sticking of H2 are carried using the static surface approximation, which suggest that we might not be able to model this slow channel. We do however make a direct comparisson to the experimental effective barrier heights obtained by applying the principle of detailed balance and direct inversion of time-of-flight measurements reported by Kaufmann et al. 45 for the fast channel.

The highly accurate potential energy surface (PES) used in our calculations and our previous work 46 has been constructed using the corrugation reducing procedure (CRP) 59 together with the SRP48 SRP density functional 32 , which was proven to be chemically accurate for H2 dissociating on Cu(111) 35 . It has also been shown previously that the SRP functional for H2 + Cu(111) is transferable to H2 + Cu(100) 30 . All our calculations have been carried out using the BOSS model which works well for activated H2 dissociation on metals at low surface temperatures 26,33–36,60 .

This paper is organized as follows. Section 2 will outline the computational methods used, with sections 2.A detailing the coordinate system used and section 2.B describes the AIMD calculations. Section 2.C and 2.D describe the QCT and QD methods respectively, while section 2.E describes the calculations of observables. Section 3 is the results and discussion section. Section 3.A is a comparison between QD and QCT reaction probabilities at the fully-state-resolved level. Sections 3.B presents calculated rotational quadrupole alignment parameters for both H2 and D2 . In section 3.C we compare theory to the experimental ef-

5

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

fective barrier heights reported by Kaufmann et al. 45 . Section 3.D presents a comparison of BOMD, QCT and molecular dynamics with electronic friction (MDEF) calculations for D2 in order to highlight the extent to which surface atom motion and electron hole-pair excitation can be expected to affect the reaction probability in molecular beam experiments. In section 3.E we present fully quantum dynamical molecular beam simulations for H2 reacting on Cu(211), comparing to QCT calculations. Section 4 presents conclusions.

6

ACS Paragon Plus Environment

Page 6 of 51

Page 7 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

2. Computational methods and simulations In the following, we present details about the different simulations we have performed to describe the dynamics of H2 (D2 ) incident on Cu(211). In our six-dimensional QD, QCT and MDEF simulations we used the static surface approximation. They are carried out on a six-dimensional PES that was previously developed by us 46 on the basis of the corrugation reducing procedure 59 and ∼116 000 DFT energy points computed with the SRP48 functional 32 . The SRP48 functional contains 48% RPBE 61 and 52% PBE 62 exchange correlation and was fitted to quantitatively reproduce experimental sticking probabilities for the reaction of H2 (D2 ) on a flat Cu(111) surface 32 . The very similar SRP functional 35 performed excellent at describing the H2 + Cu(100) reaction 63 .

2.A Coordinate system The six-dimensional dynamics calculations account only for the motion along the six molecular degrees of freedom (DOF) of H2 (D2 ) while the surface atoms are kept frozen at their ideal 0 K configuration as computed with DFT. The molecular coordinates include the center of mass (COM) position given by the coordinates X, Y, Z, where Z is molecule-surface distance and X, Y are the lateral positions measured relative to a Cu reference atom at the step edge. Also included are the H-H bond distance r and the angular orientation of H2 given by the polar angle θ defined with respect to the surface normal and the azimuthal angle φ. The coordinate system is drawn in figure 1a, and the Cu(211) surface unit cell in figure 1b, and additional details about the dimensions of the (1×1)Cu(211) unit cell are specified in the corresponding caption.

2.B Ab initio molecular dynamics simulations To describe the reaction of D2 on Cu(211) at normal incidence with the BOMD technique, we employ a modified version of the Vienna Ab Initio Simulation Package 64–67 (VASP). Note

7

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 1: Coordinate system for H2 (D2 ) on Cu(211). H atoms are drawn in blue and Cu atoms in brown. Shown in a) is a side view on a (1×2)Cu(211) supercell and b) a top view on (1 × 1) unit cell. The six-dynamical molecular DOF are indicated, i.e., the COM coordinates given by X, Y, Z, where X, Y are the lateral coordinates and Z is the molecule surface distance. Further, the H-H bond distance is represented by r and the angular orientation by the polar angle θ and the azimuthal angle φ. The latter is defined with respect to the X axis, the former with respect to the macroscopic surface normal. The computed lengths of the lattice vectors of the (1 × 1) unit cell are LX = 6.373 ˚ A and LY = 2.602 ˚ A along X and Y. that in previous publications, we referred to the direct dynamics technique using SRP-DFT as the ab-initio molecular dynamics (AIMD) method. Because this might be taken to imply that the SRP functional is not semi-empirical we abandoned this name and we now refer to it as Born-Oppenheimer molecular dynamics (BOMD). The modifications of the computer package concern the propagation algorithm and were first introduced in work 68,69 on electronically non-adiabatic effects in gas-surface systems using VASP. To be consistent with our previous work on the system 46 , we adopt the same computational setup for the electronic structure calculations specified in the supporting information of ref. 46 Here, we briefly recall only the most important details. The Cu(211) surface is represented using a five layer slab model periodically repeated over a (1×2) supercell with a vacuum spacing of 15 ˚ A. Ultrasoft pseudopotentials are used and plane waves corresponding to energies of up to 370 eV. The

8

ACS Paragon Plus Environment

Page 8 of 51

Page 9 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

k-points are sampled using the Monkhorst grid scheme and an 8×8×1 mesh centered at the Γ point. Fermi smearing is used with a width of 0.1 eV.

BOMD simulations are performed at different average incidence energies and mimic corresponding molecular beam conditions at which Michelsen et al. 70 originally performed experiments on the dissociation of D2 on flat Cu(111). The inclusion of beam parameters in the simulations is explained below in section . For each incidence energy point, we perform 500 trajectory calculations. This allows us to achieve an absolute standard error of smaller than 0.02 in the computed initial sticking coefficient. All BOMD trajectories start at a moleculesurface distance of Z = 7 ˚ A and are propagated until dissociation or scattering of D2 has occurred. Here, we count trajectories to be dissociatively adsorbed if the D-D bond distance r is larger than 2.45 ˚ A. A non-reactive scattering event is counted when trajectories return to the gas phase and have reached a molecule-surface distance of Z ≥ 7.1 ˚ A. We use a time-step discretization ∆t of 1 fs in the dynamics propagation and a maximum propagation time tf of 2 ps. Geometries between consecutive time steps are updated if the electronic structure energy is converged to 10−5 eV. The setup allows on average for an energy conservation error of typically ∼ 10 meV.

BOMD simulations performed within the static surface approximation employ the same slab model described in our earlier work 46 . Therein, the first four layers of the slab are relaxed through energy minimization (the positions of the fifth layer atoms are fixed during relaxation). The resulting optimized Cu(211) surface conserves the p1m1 space group and remains unchanged during the BOMD simulations. This prevents energy transfer to take place between the molecule and the surface due to excitation of surface atom motion upon scattering. To model a thermalized Cu(211) surface at a temperature Ts = 120 K according to experiments, we follow the NVE/NVT procedure explained in Refs. 32,71 and generate 10,000 slab configurations resembling the phase space. The initial condition of an BOMD

9

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

trajectory at Ts = 120 K is set up by randomly mixing thermalized slab models with a random configurations of D2 generated according to the molecular beam conditions.

2.C Quasi-classical simulations The MD(EF) simulations presented in this work use the 6D-PES of Ref. 46 and assume quasi-classical conditions 72 , that is, initial conditions of the classical trajectories reflect the quantum mechanical energies of incident H2 (D2 ) in their initial rovibrational state(s). To do so, we use the method described in ref. 71 The dynamics is studied by integrating a Langevin equation 73 numerically using the stochastic Ermak-Buckholz algorithm 74 and the methodology is outlined in refs. 71,75 Note that in the non-dissipative limit, i.e. the MD case, the Langevin equation obeys Newton’s equation of motion for which the propagation algorithm is also suitable. In the MDEF case, energy dissipation between molecule and surface is mediated through electronic friction as computed from the local density friction approximation within the independent atom approximation (LDFA-IAA) model 76 . Specifically, friction coefficients of the hydrogen atoms are represented as a function of the electron density of the ideal bare Cu(211) surface. The latter is extracted from a single DFT calculation, see Ref. 71 for details.

QCT calculations are used here i) to model fictitious molecular beam experiments using realistic beam parameters, and ii) to perform initial state-resolved calculations. In the former case, 50 000 QCT calculations per energy point are computed, whereas state-resolved sticking coefficients are evaluated per energy point over 5000 trajectories. As with BOMD, all MD(EF) trajectories start at a molecule-surface distance of Z = 7 ˚ A. A time step of ∆t = 0.5 h ¯ /Eh (≈0.012 fs) is used for the propagation resulting in an energy conservation error for the MD simulations of smaller than 1 meV. To determine dissociative adsorption and non-reactive scattering, we impose the same conditions used for the BOMD simulations, see above. 10

ACS Paragon Plus Environment

Page 10 of 51

Page 11 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

2.D Quantum dynamics simulations To perform 6D quantum dynamics simulations, we solve the time-dependent Schr¨odinger equation: i¯h

dΨ(Q; t) = H(Q)Ψ(Q; t), dt

(1)

using the time-dependent wave packet (TDWP) approach as implemented in our in-house computer package 56,57 . In Eq.1, Q = (X, Y, Z, r, θ, φ)T is a six-dimensional position vector, ˆ ψ(Q; t) is the time-dependent nuclear wave function of the system and H(Q) is the timeindependent Hamiltonian which reads: h ¯2 2 h ¯2 ∂2 1 ˆ2 ˆ =− ∇ − + J (θ, φ) + V (Q). H(Q) 2 2M 2µ ∂r 2µr2

(2)

Here, M and µ are the mass and the reduced mass of H2 , and ∇ and Jˆ are the Nabla and the angular momentum operators. The 6D PES, V (Q) = V (X, Y, Z, r, θ, φ), is taken from ref 46 and was computed with the SRP48 functional 32 . The initial wave function is represented as a product of a Gaussian wave packet u(Z0 , k0Z ) centered around Z0 , a two-dimensional plane wave function φ(k0X , k0Y ) along X,Y and the rovibrational wave function ψν,j,mj (r, θ, φ) of incident H2 : Ψ(Q, t = 0) = ψν,j,mj (r, θ, φ)φ(k0X , k0Y )u(Z; Z0 , k0Z )

(3)

where the two-dimensional plain wave function and the Gaussian wave packet are defined as X

Y

φ(k0X , k0Y ) = ei(k0 X0 +k0 Y0 ) u(Z; Z0 , k0Z ) =



2σ π

2 1 Z 4

0



dk0Z e−σ

(4) 2 (k−k Z ) 0

Z

Z

ei(k−k0 )Z0 eik0 Z0 .

(5)

Here, σ is the width of the wave packet centered around the wave vector k and k0X,Y,Z are the initial wave vectors of the COM. The width σ is chosen in such a way that 90% of the Gaussian wave packet is place in a energy range Ei ∈ [Emin , Emax ]. Eq.1 is solved numer-

11

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 51

ically using the split operator method with a time step ∆t. We apply a quadratic form of optical potentials 77 in the scattering (large values of Z) and adsorption regions (large values of r). The scattered fraction of the wave function is analyzed through the scattering matrix formalism 78 and the scattering probability Psc is computed accordingly. Substracting Psc from 1 then yields the sticking probability S0 .

Parameters for the wave packet calculations defining the initial wave packet, grid representation, time step and the optical potentials are compiled in Table 1. The final propagation time can vary since we stop simulations if the remaining norm on the grid is below 0.01. Table 1: Input parameters for the 6D quantum simulations on the reactive scattering of H2 on Cu(211). All wave packets were propagated until the remaining norm was less then one percent. 0.05 - 0.22 eV ν0 ν1 Zstart (Bohr) -2.0 -2.0 NZspec 280 280 NZ 180 180 ∆Z (Bohr) 0.1 0.1 Rstart (Bohr) 0.8 0.8 NR 60 60 ∆R (Bohr) 0.15 0.15 NX 36 36 NY 12 12 NJ 26 / 25 30 / 29 NmJ 26 / 25 30 / 29 Complex absorbing potentials Z CAP start [a0 ] 8.9 8.9 Z CAP end [a0 ] 15.9 15.9 Z CAP Optimum [eV] 0.16 0.16 CAP start [a ] Zspec 18.1 18.1 0 CAP end [a ] Zspec 25.9 25.9 0 CAP Optimum [eV] Zspec 0.16 0.16 RCAP start [a0 ] 4.55 4.55 RCAP end [a0 ] 9.65 9.65 RCAP optimum [eV] 0.12 0.12 Propagation ∆t [¯ h/Eh ] 2 2 tf [¯ h/Eh ] 44000 44000 Initial wave packet Emin [eV] 0.05 0.05 Emax [eV] 0.22 0.22 Z0 [a0 ] 13.50 13.5

0.2 - 0.6 eV ν0 ν1 -2.0 -2.0 280 280 176 176 0.08 0.08 0.8 0.8 56 56 0.15 0.15 36 36 12 12 26 / 25 32 / 31 26 / 25 32 / 31

ν0 J ∈ [0, 7] -2.0 280 176 0.08 0.8 56 0.15 36 12 38 / 37 30 / 29

0.57 - 1.4 eV ν0 J ∈ [8, 11] -2.0 280 176 0.08 0.8 56 0.15 36 12 42 / 41 42 / 41

ν1 -2.0 280 176 0.08 0.8 56 0.15 36 12 36 / 35 28 / 27

D2 ν1 J6 -2.0 280 176 0.08 0.8 56 0.15 42 16 42 40

8.88 12.0 0.3 16.8 20.32 0.3 4.55 9.05 0.3

8.88 12.0 0.3 16.8 20.32 0.3 4.55 9.05 0.3

8.88 12.0 0.95 18.16 20.32 1.2 4.55 9.05 1.0

8.88 12.0 0.95 18.16 20.32 1.2 4.55 9.05 1.0

8.88 12.0 0.95 18.16 20.32 1.2 4.55 9.05 1.0

8.88 12.0 0.3 16.8 20.32 0.3 4.55 9.05 0.3

2 14000

2 14000

2 10000

2 10000

2 10000

2 20000

0.2 0.6 11.44

0.2 0.6 11.44

0.57 1.4 11.44

0.57 1.4 11.44

0.57 1.4 11.44

0.2 0.6 11.44

12

ACS Paragon Plus Environment

Page 13 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

2.E Computation of observables To incorporate the effect of a molecular beam on the computed sticking coefficient we need to take into account the distributions of translational energies and rovibrational state population due to a nozzle temperature Tn . The probability to find a molecule with velocity v + dv and a rovibrational state described by the vibrational quantum number ν and the angular momentum quantum number j is here given by:

P (v, ν, j, Tn )dv = Pf lux (v; Tn )dv × Pint (ν, j, Tn )

(6)

where the flux-weighted velocity distribution Pf lux is a parameterized function of Tn and determined by the width parameter α and the stream velocity v0 according to 79 2 /α2

Pf lux (v; Tn )dv = Cv 3 e−(v−v0 )

dv

(7)

where C is a normalization constant. The ensemble representation of the rovibrational state population distribution reads: w(j)f (ν, j, Tn ) 0 0 v 0 ,j 0 ≡j(mod 2) f (ν , j , Tn )

Pint (ν, j, Tn ) = P

(8)

with f (ν, j, Tn ) = (2j + 1) × e(−(Eν,0 −E0,0 )/kB Tvib ) × e(−(Eν,j −Eν,0 )/kB Trot ) .

(9)

Here, kB is the Boltzmann constant and Eν,j is the energy of the quantum state characterized by ν and j. The first and second Boltzmann factor describe vibrational and rotational state populations, respectively. Note, that the rotational temperature is Trot = 0.8Tn 18 whereas for the vibrational temperature applies Tvib = Tn . This setting is in agreement with the observation that rotational but no vibrational cooling occurs during gas expansion in the nozzle. The factor w(j) in Eq. 8 is due to ortho- and para-hydrogen molecules present in

13

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 51

the beam. For H2 , w(j) is 1/4 (3/4) for even (odd) values of j, and for D2 , w(j) = 2/3 (1/3) for even (odd) values of j.

In the case of classical dynamics calculations (MD, MDEF and BOMD), the probability distributions P (v, ν, j, Tn ) is randomly sampled as described in ref 71 using the different beam parameters on H2 and D2 listed in Table 2. The sticking coefficient per energy point is given by the ratio of the number of adsorbed trajectories Nads and the total number of computed trajectories N , that is, S0 = Nads /N . To extract quantum mechanical results on H2 beam simulations, a direct sampling of P (v, ν, j, Tn ) is not feasible. Instead, initial state-resolved reaction probabilities Rmono (Ei , ν, j) are first computed as functions of the monochromatic incidence energy Ei by degeneracy averaging fully initial state-resolved reaction probabilities PR (Ei , ν, j, mj ) over the magnetic rotational quantum number mj , i.e., j X

Rmono (Ei , ν, j) =

(2 − δmj ,0 )PR (Ei , ν, j, mj )/(2j + 1).

(10)

mj =0

The initial sticking probability S0 (hEi i) is then calculated as a function of average incidence energy hEi i by averaging over the rovibrational (ν, j) states populated in the beam (see Eq. 8) and the flux-weighted distribution of the incidence translational energies of the beam, according to R∞

S0 (hEi i) =

XX j

ν

0

P 0 (Ei , ν, j, Tn )Rmono (Ei , ν, j)dEi R∞ . 0 0 P (Ei , ν, j, Tn )dEi

(11)

We note that although S0 (hEi i) is written and plotted in publications as a function of average incidence only, it also implicitly depends on Tn through the distribution P 0 (Ei , ν, j, Tn ) of incidence energies and the rovibrational state populations

P 0 (Ei , ν, j, Tn )dEi = Pf0 lux (Ei ; Tn )dEi × Pint (ν, j, Tn ).

(12)

P 0 (Ei , ν, j, Tn ) makes the initial sticking also depend implicitly on incident beam conditions 14

ACS Paragon Plus Environment

Page 15 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

other than just Tn , due to the occurrence of the flux-weighted distribution of incidence energies Pf0 lux (Ei ; Tn ), which depends on a number of factors including the molecular beam geometry, backing pressure, whether or not a seeding gas is used, and can be described by the parameters E0 and ∆E0 according to: √

Pf0 lux (Ei ; Tn )dEi = C 0 Ei e−4E0 (

√ Ei − E0 )2 /∆E02

dEi .

(13)

Instead of averaging over incidence energies using Pf0 lux (Ei ; Tn ) as done in Eq. 12 it is also possible to average over the flux-weighted velocity distribution of the molecules in the beam, Pf0 lux (vi ; Tn ), and the derivation Pf0 lux (Ei ; Tn ) from Pf lux (v; Tn ) is discussed in Ref 79 . For a particle of mass m, the parameters are defined as E0 = mv02 /2 and ∆E0 = 2E0 α/v0 .

To obtain sticking coefficients S0 , we perform 114 state-resolved calculations (corresponding to 342 wave packet calculations) for an energy range of Ei ∈ [0.05, 1.4] eV. The initial states of incident H2 considered here to evaluate Eq.11 are characterized by the quantum numbers J ∈ [0, 11] for ν = 0 and J ∈ [0, 7] for ν = 1, respectively, and mJ ∈ [0, J].

The rotational quadrupole alignment parameter as a function of ν and j is a measure of the extent to which the reaction depends on the orientation of the molecule. The rotational quadrupole alignment parameter is calculated from the fully-state-resolved reaction probability as follows 80 

Pj (2)

A0 (ν, j) =

mj =0 (2 − δmj ,0 )Pr (ν, j, mj )

Pj

mj =0 (2

3m2j j(j+1)

− δmj ,0 )Pr (ν, j, mj )

15

ACS Paragon Plus Environment



−1

.

(14)

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 51

Table 2: Molecular beam parameters taken from experiments performed on the H2 (D2 ) + Cu(111) system. The parameters are used in this work to simulate the reaction of molecular hydrogen on Cu(211) as it would occur in experiments analogous to those performed on Cu(111). The parameters v0 , α, Tn represent the stream velocity of the beam, the width of the beam and the nozzle temperature at an average translational incidence energy hEi i. Parameters were taken from refs 15,34 Tn [K] hEi i [kJ/mol] v0 [m/s] E0 [eV] Seeded molecular H2 beams, TS = 120K 1740 19.9 3923 0.160 1740 28.1 4892 0.250 38.0 5906 0.364 1740 2000 18.2 3857 0.155 25.1 4625 0.223 2000 44.1 6431 0.432 2000 Seeded molecular D2 beams, TS = 120K 2100 62.6 5377 0.829 69.2 5658 0.860 2100 2100 80.1 6132 0.849 Pure molecular H2 beam, TS = 120K 1435 31.7 5417 0.307 1465 32.0 5446 0.310 1740 38.0 5906 0.364 1855 40.5 6139 0.394 2000 44.1 6431 0.432 2100 47.4 6674 0.465 2300 49.7 6590 0.454 Pure molecular H2 beam, Rendulic and co. 1118.07 25.1 3500 0.12794 1331.89 29.9 3555 0.13200 1438.82 32.3 3380 0.11932 1501.19 35.7 3151 0.10371 1581.35 35.5 3219 0.10816

16

ACS Paragon Plus Environment

α [m/s] 1105 1105 945 995 1032 886 649 717 830 826 830 945 899 886 913 1351 1996 2342 2611 2819 2903

Page 17 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

3. Results and discussion 3.A Fully-state-resolved reaction probabilities In order to highlight the difference between a QD and QCT treatment of the H2 + Cu(211) system we first present initial-state-resolved reaction probabilities in figure 2a, 2b and 2c. QD calculations have been performed for a large number of rovibrational states. All input parameters can be found in table 1. The biggest differences between QD and QCT calculations at the fully-state-resolved level are observed for the lowest rovibrational states, as shown in figure 2b and 2c. The differences get increasingly smaller with increasing J for J > 1. From QCT data at higher translational energies that are not shown in this figure it is clear that all states converge towards an asymptotic maximum reaction probability which depends slightly on the rovibrational state with respect to the maximum reaction probability. We note that for very high J, J > 10 (not shown here), QD predicts a marginally smaller (less then 2%) asymptotic maximum reaction probability, while figure 2c suggest the opposite is true for the vibrational ground state and the first vibrationally excited state.

Figure 2b shows the largest discrepancy between the QCT and QD calculations observed. Here |mJ | = J pertains to a ’helicoptering’ H2 molecule, and mJ = 0 to a ’cartwheeling’ H2 molecule rotating in a plane perpendicular to the surface normal. The preference for reacting parallel to the surface (i.e. mJ = J having a higher reaction probability then mJ = 0) is bigger for QD calculations than for QCT calculations. This difference is negligible however when looking at degeneracy averaged reaction probabilities, which are shown in figure 2a. This also holds for the states not shown here. When looking at degeneracy averaged reaction probabilities, the agreement between the QCT and QD method is excellent.

In our calculations we see no evidence of the ”slow channel” reactivity reported by Kaufmann et al. 45 in their very recent paper, i.e. of reaction at low translational energies. We can now rule out quantum effects during the dynamics as the source of this slow channel 17

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 51

reactivity, in which reaction supposedly is inhibited by translational- and promoted by vibrational energy 45 . When looking at the individual rovibrational states that exhibit the biggest difference in reactivity between QD and QCT calculations, no evidence of the slow reaction channel is present in our results. The translational energy range sampled in our calculations should overlap with the translational energy range where the slow channel is reported to be active by Kaufmann et al. 45 . We therefore propose that the observed slow reaction channel must originate from surface motion at a very high surface temperature (923K), which has not been incorporated into our QD calculations and is challenging to incorporate in QCT calculations 81 .

3.B Rotational quadrupole alignment parameters As might be suspected from figure 2b from the larger preference for a parallel reaction orientation for J = 1, calculated rotational quadrupole alignment parameters show a large difference between QCT and QD calculations for the J = 1 states shown there. However, here we will now focus on rotational quadrupole alignment parameters for two particular rovibrational states of H2 : (ν = 0, J = 7) and (ν = 1, J = 4) (figure 3a), and D2 : (ν = 0, J = 11) and (ν = 1, J = 4) (figure 3b). These two sets of states were selected because they are very similar in rotational energy to the two rovibrational states for which rotational quadrupole alignment parameters for D2 desorbing from Cu(111) have been measured experimentally 16 and studied theoretically using the BOMD method 32 . Results for both states of D2 react(2)

ing on Cu(111) have been included in figure 3b. Note that a positive A0 (ν, J) indicates a preference for a parallel reaction orientation, a negative value indicates a preference for a perpendicular orientation, and zero means the reaction proceeds independent of orientation.

We observe that the predicted rotational quadrupole alignment parameters eventually tend to zero with increasing translational energy, as all molecules irrespective of orientation will

18

ACS Paragon Plus Environment

Page 19 of 51

0.8

ν0J1

Degeneracy averaged J=1

ν1J1

0.6 0.4 0.2

reaction probability

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

(a)

0.8

ν0J1mJ0

0.6

ν1J1mJ0

J=1

ν0J1mJ1 ν1J1mJ1

0.4 0.2 (b)

0.8

ν0J0mJ0

J=0

ν1J0mJ0

0.6 0.4 0.2 (c)

0 0.2

0.4

0.6 0.8 1 1.2 translational energy [eV]

1.4

Figure 2: Reaction probability computed with QD calculations (solid lines) and QCT calculations (dashed lines) for normal incidence. Panel (a) shows degeneracy averaged reaction probabilities for J = 1 for both the ground state and the first vibrationally excited state. Panel (b) shows the mJ = 0, 1 states belonging to J = 1 for both the ground state and the first vibrationally excited state. Panel (c) shows the J = 0 state for both the ground state and the first vibrationally excited state as well.

19

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

have enough energy to traverse the barrier. It is also clear that for H2 (ν = 0, J = 7) the agreement between QCT and QD calculations is excellent. The slight deviations at the lowest translational energies can be attributed to noise in the very low reaction probabilities of the underlying individual states.

The increase of the rotational quadrupole alignment parameter with decreasing translational energy, for the H2 (ν = 0, J = 7) and D2 (ν = 0, J = 11) states, is comparable to what is reported in the literature for H2 and D2 associatively desorbing from Cu(111) and Cu(100) 16,30–32 . This monotonic increase of the rotational quadrupole alignment parameter with decreasing translational energy can be explained by a static effect of orientational hindering, in which slow- or non-rotating molecules will scatter when their initial orientation does not conform to the lowest barrier geometry 31 . Specifically, the molecule must be in favourable orientation to begin with in order to react, especially with the energy available to reaction being close to the threshold energy.

The blue lines in figure 3a correspond to the (ν = 1, J = 4) rovibrational state of H2 and in figure 3b to the (ν = 1, J = 6) rovibrational state of D2 . In contrast to the previously described states (ν = 0,high J), the rotational quadrupole alignment parameter now first increases with increasing translational energy until reaching a maximum around 0.43 eV for H2 and 0.52 eV for D2 before decreasing towards zero with increasing translational energy. From figure 3 it is clear that around the maximum the agreement between the QD and QCT calculations is not as excellent for H2 than for D2 , although the agreement is still good.

The downturn of the rotational quadrupole alignment parameter with decreasing translational energy seen here for D2 and H2 in their (ν = 1) states colliding with Cu(211) was not observed for D2 desorbing from Cu(111) for which, as can be seen in figure 3b, only a monotonous increase with decreasing translational energy has been reported 16,32 . A slight

20

ACS Paragon Plus Environment

Page 20 of 51

Page 21 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

downturn of the rotational quadrupole alignment parameters has been predicted for vibrationally excited H2 reacting on Cu(100) 30,31 , although the downturn was too small to lead to a negative rotational quadrupole alignment parameter. Because the behavior predicted for (ν = 1) hydrogen colliding with Cu(211) qualitatively differs from that observed previously for Cu(111) and Cu(100), we will now first attempt an explanation for the dependence of the rotational quadrupole alignment parameter on incidence energy that we predict for D2 (ν = 1, J = 6), and then discuss the case D2 (ν = 0, J = 11).

From the literature it is known that the behavior of the rotational quadrupole alignment parameter as a function of incidence energy can be related to features of the molecule-surface interaction at the preferred reaction site of the molecule, for the initial rovibrational state considered 31 . For example, vibrationally excited H2 with a translational energy close to the threshold to reaction was found to prefer to react on a top site of Cu(100) due to features in the PES being more favorable, for instance, the increased lateness of the barrier at this site allowed more efficient conversion of energy from vibration to motion along the reaction path 31,37,38 . Next, the dependence of the rotational quadrupole alignment parameter on incidence energy of vibrationally excited H2 on Cu(100) could be explained on the basis of the anisotropy of the molecule-surface interaction energy at the top site. In our explanation of the behavior seen for H2 and D2 on Cu(211), we will therefore proceed in a similar manner.

Figure 4a shows the reaction density of D2 (ν = 1, J = 6) extracted from QCT calculations projected onto the Cu(211) unit cell. Here we focus specifically on the D2 (ν = 1, J = 6) rovibrational state because it has been experimentally measured on Cu(111) 16 , but the same mechanism appears to be present in our data for H2 (ν = 1, J > 2). All reacted trajectories up to a translational energy of 0.35 eV have been included. It is immediately clear that molecules in this particular state prefer to react on the t1 top site 46 , which in the case of Cu(211) is at the step, with small outliers in reactivity pointing towards the bottom of the

21

ACS Paragon Plus Environment

The Journal of Physical Chemistry

(2)

1 rotational quadrupole alignment parameter A0(ν,J)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

H2

Page 22 of 51

(2)

A0(0,7) A(2) (1,4) 0

0.8 0.6 0.4 0.2 0

(a)

-0.2 0.2

0.4

0.6

0.8

2

1

1.2

(2)

A0(0,11)

D2

1.5

(2)

A0(1,6) (2)

A0(1,6) exp. Cu(111) (2)

A0(1,6) BOMD Cu(111)

1 0.5 0

(b)

-0.5 0

0.2

0.4 0.6 0.8 1 translational energy [eV]

1.2

(2)

Figure 3: Panel (a) shows rotational quadrupole alignment parameters, A 0 (ν, J), for two rovibrational states of H2 : (ν = 0, J = 7) and (ν = 1, J = 4). Panel (b) shows rotational quadrupole alignment parameters for two rovibrational states of D2 : (ν = 0, J = 11) and (ν = 1, J = 6). Solid lines correspond to QD calculations, dashed lines to QCT calculations. Panel (b) also shows experimental results for D2 on Cu(111) (black) 16 , as well as BOMD results for D2 (ν = 1, J = 6) on Cu(111) (green) 32 .

22

ACS Paragon Plus Environment

Page 23 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

step. The t1 barrier is an extremely late barrier (rt1 = 1.44 ˚ A), as can be seen in table 3 of ref 46 . The very late barrier allows for efficient conversion of vibrational energy to motion along the reaction coordinate 10,11,40 .

Figure 5 shows a representative reactive trajectory of D2 (ν = 1, J = 6, mJ = 0) with a translational energy of 0.3 eV, and plots the classical angular momentum, JC , as a function of the propagation time. JC is decreased before reaching the barrier, and a minimum in JC is reached at the transition state, where r becomes equal to 1.44 ˚ A corresponding to the t1 barrier 46 . In the majority of the reacted trajectories the minimum of JC is reached when r reaches the value of the t1 transition state, even when the molecule would make one or more bounces on the surface. This is a clear indication that rotational de-excitation takes place before the molecule reaches the transition state. This suggests that the reaction proceeds through rotational inelastic enhancement 31 , i.e., the reaction is promoted by rotational energy flowing to the reaction coordinate. The bump in JC (i.e. its increase) still relatively far away from the surface is a feature that is also present in the majority of reactive trajectories. It is not completely clear to us what the cause is of this increase of JC still relatively far away from the surface before proceeding towards the transition state. We speculate that the increasing vicinity to the surface turns on the anisotropy of the molecule-surface interaction, thereby coupling rotational motion and stretching motion, and providing a mechanism for the rotational energy to remain more constant while the bond extends and compresses due to the molecular vibration. This mechanism could consist in the classical angular momentum increasing when the bond extends, to offset the effect of the bond extension on the rotational constant (upon bond extension the rotational constant decreases and if not compensated this would decrease the rotational energy). This could possibly explain the hump observed in JC at t ≈ 4500 atomic units of time in figure 5.

There is also indirect evidence for rotationally enhanced reaction of D2 (ν = 1, J = 6, mJ =

23

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

0) in our QD calculations. Figure 6a shows inelastic scattering probabilities for D2 (ν = 1, J = 6, mJ = 0) and figure 6b shows inelastic scattering probabilities for D2 (ν = 1, J = 6, mJ = 6). From a pair-wise comparison of data with the same color between figure 6a and figure 6b it is clear that D2 (ν = 1, J = 6, mJ = 0) has a considerably higher probability to rotationally de-excite in the scattering process compared to D2 (ν = 1, J = 6, mJ = 6). This suggests that the reaction of (ν = 1, J = 6, mJ = 0) is also rotationally enhanced in the quantum dynamics if the de-excitation occurs before the barrier is reached and the released rotational energy is transfered to motion along the reaction coordinate.

There are four possible mechanisms that affect the reaction probability and may affect the rotational quadrupole alignment parameters, two enhancing mechanisms and two steric hindering mechanisms 31 . Here we have focused on one enhancement mechanism, inelastic rotational enhancement, since the evidence presented in figures 5, 6a and 6b is consistent with this mechanism. Inelastic rotational enhancement requires reaction to take place on a site with a low anisotropy in φ and a large anisotropy in θ at the barrier 31 . The main reasons for proposing the presence of this mechanism are the sharp downturn of the quadrupole alignment parameters for (ν = 1, J > 2) rovibrational states in figure 3a and 3b and the rotational de-excitation seen in figures 5, 6a and 6b. We note that inelastic rotational enhancement is the only mechanism that predicts a lowering of the rotational quadrupole alignment parameters 31 . A complete overview of the four mechanisms and what features of the PES they depend on can be found in table 3 of ref. 31 .

A feature of the t1 site that facilitates the conversion of rotational energy to motion along the reaction coordinate is a low anisotropy of the potential in φ combined with a large anisotropy in θ. Figure 7 shows the anisotropy at the t1 barrier 46 (r and Z are kept constant here), the top panel shows the anisotropy in φ and the bottom panel shows the anisotropy in θ. It is clear that the anisotropy in θ is substantial, while the anisotropy in φ is very small com-

24

ACS Paragon Plus Environment

Page 24 of 51

Page 25 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

pared to the anisotropy in θ. Somers et al. 31 have shown that the high anisotropy in θ may facilitate inelastic rotational enhancement. Inelastic rotational enhancement is expected to be most effective for low |mJ | states with J > 2, and the mechanism would lead to decreased rotational quadrupole alignment parameters 31 . The reason for the decrease in the rotational quadrupole alignment parameters is that mJ is approximately conserved, so that a decrease in J is possible only for low |mJ |.

It is also clear from figure 3b that from the point of view of the orientational dependence of reaction Cu(211) cannot be described as a combination of (100) steps and (111) terraces. The monotonic increase of the rotational quadrupole alignment parameter for D2 reacting on Cu(111) 16,32 is very similar to the behavior reported for Cu(100) 30,31 . A slight downturn at translational energies close to the threshold to reaction has been reported in the case of Cu(100), indicating that the inelastic rotational enhancement mechanism is taking place. The downturn is however small and does not lead to negative quadrupole alignment parameters as we show here for H2 and D2 reacting on Cu(211). This is a clear indication that the reaction dynamics of the Cu(211) surface are distinct from the reaction dynamics of its component Cu(111) terraces and Cu(100) steps when looked at individually. This is most likely because the energetic corrugation of the Cu(211) surface is much lower compared to Cu(111) and Cu(100), a feature that favors the reaction of vibrationally excited molecules if sites with late barriers are present.

We now turn to an explanation for the monotonic decrease of the rotational quadrupole alignment parameter predicted for the (ν = 0, high J) states of H2 and D2 colliding with Cu(211) in figure 3a and 3b. No downturn of the rotational quadrupole alignment parameter is observed for the (ν = 0) states even though D2 (ν = 0, J = 11, mJ = 11) reacts at the step as well as D2 (ν = 1, J = 6), as can be seen in figure 4b. The lack of a downturn in the rotational quadrupole alignment parameter arises because the D2 (ν = 0) states re-

25

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

act using a different mechanism. Figure 8 shows a representative reactive trajectory of D2 (ν = 0, J = 11, mJ = 11), and it is clear that the angular momentum only drops after the transition state has been reached. This is a clear combination of elastic rotational enhancement for the helicopter molecules together with orientational hindering for the cartwheeling molecules which causes the increase of the rotational quadrupole alignment parameters of the (ν = 0) molecules 31 . We note that D2 (ν = 0, J = 11) reacting on the step at the t1 site is due to the high initial rotational quantum number. The t1 barrier is slightly higher in energy than the lowest barrier to reaction but at this site the reaction is less rotational hindered if the molecule rotates in a plane parallel to the surface, and the barrier is much later than at the lowest b2 barrier on the terrace. This allows molecules in the vibrational ground state that are rotating fast in helicopter fashion and have incidence energies close to the threshold to reaction to react there, by converting rotational energy to motion along the reaction path as the bond extends and the rotational constant of the molecule drops, while j remains roughly the same.

Above, we have shown that D2 in its (ν = 0, J = 11) and (ν = 1, J = 6) states prefers to react near the t1 -site, i.e., on or near the steps (see figure 4a and 4b). This might seem to contradict an earlier conclusion, that at low incidence energies D2 prefers to react on the terrace 46 . However, this conclusion was based on molecular beam experiments and simulations of those experiments, and under the conditions addressed 46 the (ν = 0, J = 11) and (ν = 1, J = 6) states would hardly have population in them. A more appropriate picture of the reaction density for molecules under the conditions of Ref. 46 is shown in figure 4c. There it can be seen that D2 (ν = 0, J = 2) (this state would be highly populated in the beams used and simulated in Ref. 46 ) prefers to react at the terrace b2 site, which has the lowest barrier to reaction. The reaction density for D2 (ν = 0, J = 2) is in line with earlier findings that molecules in the vibrational ground state with low j react at the lowest barrier to reaction 31,37,38 , and with the findings for D2 + Cu(211) of Ref. 46 .

26

ACS Paragon Plus Environment

Page 26 of 51

Page 27 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Kaufmann et al. 45 did not measure rotational quadrupole alignment parameters in their recent study. We believe that the downturn of the rotational quadrupole alignment parameter at low incidence energies, which has not been observed before with this large downward shift for both H2 and D2 reacting on copper, may well be experimentally verified for both isotopes on Cu(211). Specifically, the reaction probability of H2 and D2 is large enough, and the (ν = 1, J = 4) rovibrational state of H2 and the (ν = 1, J = 6) of D2 have large enough Boltzmann weights at reasonable surface temperatures (923K) to make the downturn measurable. Comparing experimental rotational quadrupole alignment parameters to theoretical ones will provide a very stringent, and detailed way of testing the accuracy of the electronic structure calculations used in the construction of the PES.

3.C Comparing to experimental E0 (ν, J) parameters Next we will make a direct comparison with the state-specific, or degeneracy averaged, reaction probabilities reported by Kaufmann et al. 45 From their experiments they could derive dissociative adsorption probabilities by applying the principle of detailed balance to the measured time-of-flight distributions. However, comparing the relative saturation value of the reaction probability obtained from associative desorption experiments to the zero coverage absolute saturation values predicted by theory is not straightforward. The authors of the experimental paper pose several ways of scaling the experimental data in order to make a comparison to theoretical work possible. Scaling the experimental data to experimental molecular adsorption results introduces the uncertainties related to the direct molecular adsorption experiment used as a reference in this process. Theory calculates sticking probabilities in the zero coverage limit. When scaling the experimental desorption data to experimental adsorption data the zero coverage limit will only be a lower bound, especially when a molecular beam experiment with a very broad translational energy distribution is

27

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 4: Three plots of the reaction density of D2 projected onto the Cu(211) unit cell. Panel a shows the reaction density of D2 (ν = 1, J = 6), all reacted trajectories up to a translational energy of 0.35 eV are included. Panel b shows the reaction density of D2 (ν = 0, J = 11), all reacted trajectories up to a translational energy of 0.35 eV are included. Panel c shows the reaction density of D2 (ν = 0, J = 2), all reacted trajectories up to a translational energy of 0.65 eV are included.

28

ACS Paragon Plus Environment

Page 28 of 51

Page 29 of 51

8

7 6

_

6

5

4

JC

4

r [Å]

3

Z [Å]

[Å]

angular momentum [h ]

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

2

2

1 0

0

2000

4000

6000

0

a.t.

Figure 5: A single representative reactive trajectory of D2 (ν = 1, J = 6, mJ = 0) with a translational energy of 0.3 eV, blue shows the angular momentum (JC ), red shows the bond length (r), and green shows the center of mass distance to the surface (Z). chosen as a reference.

We opt for the simplest and most direct method to scale to the relative experimental associative desorption data. In order to compare to the experimental E0 (ν, J) parameters, where E0 (ν, J) is the translational energy for which the reaction probability of the (ν, J) state is half of the maximum reaction probability measured for that (ν, J) state, we use the reported maximum translational energy sensitivity reported in tables S7 and S9 of ref. 45 Theoretical E0 (ν, J) are taken to be the translational energy to which the reaction probability is half that of the reaction probability at the maximum translational energy for which the experiment is sensitive. This method also corresponds to what is showcased in figure 13a of ref. 45

Figure 9 shows E0 (ν, J) parameters for H2 and D2 reacting on Cu(211). The agreement between theory and experiment is excellent for H2 . We calculated mean absolute and mean signed deviations between the experimental and theoretical E(ν, J) parameters, see table 3. It is clear from figure 9 and table 3 that the agreement between theory and experi29

ACS Paragon Plus Environment

The Journal of Physical Chemistry

ν1J6mJ0 → ν1J0

(a)

ν1J6mJ0 → ν1J2

0.2 rotational inelastic scattering probability

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 30 of 51

ν1J6mJ0 → ν1J4

0.1

0

ν1J6mJ6 → ν1J0

(b)

0.2

ν1J6mJ6 → ν1J2 ν1J6mJ6 → ν1J4

0.1

0 0.2

0.3

0.4 0.5 translational energy [eV]

0.6

Figure 6: Rotational inelastic scattering probabilities for D2 for two different inital rovibrational states as a function of translational energy. Panel (a) shows rotational inelastic scattering probabilities for D2 (ν = 1, J = 6, mJ = 0), panel (b) shows rotational inelastic scattering probabilities for D2 (ν = 1, J = 6, mJ = 6). Colors correspond to the final rovibrational state of the molecule, with blue being (ν = 1, J = 0), red being (ν = 1, J = 2), and green being (ν = 1, J = 4).

30

ACS Paragon Plus Environment

ϕ 1.1 1 0.9 0.8 0.7

Potential energy [eV]

0.6

0

60

120

180

240

300

360

90 120 degree

150

180

θ

30 20 10 0

0

30

60

Figure 7: Anisotropy of the interaction potential at the t1 top site barrier 46 for θ (top panel) and φ (bottom panel).

7

14

6

12

5

10 8 6

JC

4

r [Å]

3

Z [Å]

4

2 1

2 0

[Å]

_

angular momentum [h ]

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Potential energy [eV]

Page 31 of 51

0

2000

4000 a.t.

6000

0 8000

Figure 8: A single representative reactive trajectory of D2 (ν = 0, J = 11, mJ = 11) with a translational energy of 0.3 eV, blue shows the angular momentum (JC ), red shows the bond length (r), and green shows the center of mass distance to the surface (Z). 31

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 32 of 51

ment is excellent in the case of H2 , for which the total mean absolute deviation (MAD) (n−1

P

n

|E0,exp − E0 |) and mean signed deviation (MSD) (n−1

P

n

E0,exp − E0 ) values for QD

and QCT calculations fall within chemical accuracy. We note that for H2 the agreement is best for vibrationally excited molecules, while the reverse is true with respect to D2 . For D2 the agreement is not yet within chemically accuracy, mainly due to the slightly bigger discrepancies between theory and experiment for the first vibrationally excited state. Theory, however, does not reproduce the rotational hindering that can be seen in the experimental data, i.e. E0 (ν, J) does not first increase with J until a maximum before falling off with increasing J. Theory shows no such behavior, here the E0 (ν, J) parameter falls off with increasing J for all methods investigated here.

Experiments on associative desorption of H2 from Cu(111) 18,45 and of D2 from Cu(111) 15,32,45 likewise found the rotational hindering effect on reaction for low j. As for H2 and D2 interacting with Cu(211), we have not been able to reproduce this subtle effect in calculations on H2 + Cu(111) 34,35 and D2 + Cu(111) 32,34 in electronically adiabatic dynamics calculations. Here we find that MDEF calculations on H2 and D2 + Cu(211) do not reproduce the trend either, suggesting that in the previous calculations the neglect of electron-hole pair excitation was not the cause of the discrepancy between theory and experiment. However, it is possible that calculations modeling electron-hole pair excitation with orbital-dependent friction (ODF) will succeed in recovering the subtle trend observed in experiments. For this, it may well be necessary that the ODF coefficients explicitly model the dependence of the tensor friction coefficients on the molecule’s orientation angles; earlier MDEF calculations on H2 + Cu(111) using ODF coefficients did not yet do this 25 .

According to figure 9, the reactivity measured experimentally in the associative desorption experiments is, for most (ν, J) states, larger than that predicted theoretically, with the experimental E0 (ν, J) being lower. With the use of the same scaling method to relate theory to

32

ACS Paragon Plus Environment

Page 33 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

experiment, Kaufmann et al. 45 obtained the same result for H2 and D2 reacting on Cu(111), and also in their case they compared with theory on the SRP48 functional 32 . To some extent these results are odd, as calculations for H2 and D2 + Cu(111) using the original SRP functional showed that the theory overestimated the experimentally measured sticking coefficients 35 . However, also in this work theory generally underestimated the reactivity measured in associative desorption experiments 35 .

The paradox noted above may be explained on the basis of the BOSS model used in the calculations. This model neglects the effect of ehp excitation. Modeling this effect on sticking experiments should lower the theoretical reactivity, with computed sticking curves shifting to higher energies. Modeling the effect on associative desorption experiments should show the opposite effect, if the modeling is done correctly, i.e., starting with molecules being formed at the transition state and then desorbing 35,82 . The effect of ehp excitation in such calculations should lead to translational energy distributions of desorbed molecules being shifted to lower translational energies. The reaction probability curves obtainable from these distributions by assuming detailed balance (which, strictly speaking, is not applicable if ehp excitation is active) should then lead to computed reaction probability curves (E0 (ν, J) values) shifted towards lower energies, in better agreement with experiment (see figure 9).

The above also explains why our present MDEF calculations led to decreased agreement with experiment: In these calculations we modeled the associative desorption experiment as an initial-state selected dissociative chemisorption experiment, in which ehp excitation should have the opposite effect. If we assume the ehp excitation to have an effect that is similar in magnitude, but opposite in sign with respect to the QCT calculations, the net effect of modeling ehp excitation is to increase the agreement with experiment to the extent that chemical accuracy is obtained for both (ν = 0) and (ν = 1) H2 on Cu(211). This is illustrated by the MDEF* mean absolute and mean signed deviations in table 3. The MDEF* values have

33

ACS Paragon Plus Environment

The Journal of Physical Chemistry

been calculated by subtracting the difference between the MDEF and QCT values from the QCT values. We finally note that we have assumed that surface temperature does not much affect the measured E0 (ν, J) through surface atom vibrational motion, which is in line with experiments 23,24 , as discussed in the supporting information of D´ıaz et al. 35 .

Table 3: Mean absolute and mean signed deviations for the theoretical E0 (ν, J) parameters compared to experimental values shown in figure 9.

QCT QD MDEF MDEF*

QCT

MAD [eV] total ν=0 0.0362 0.0384 0.0362 0.0449 0.0509 0.0531 0.0239 0.0237 MAD [eV] total ν=0 0.0485 0.0354

H2 ν = 0

0.9

H2 ν=1 0.0289 0.0235 0.0272 0.0306 D2 ν=1 0.0675

H2 ν = 1

MSD [eV] total ν=0 0.0209 0.0384 0.0241 0.0449 0.0342 0.0532 0.0076 0.0236 MSD [eV] total ν=0 0.0485 0.0354

D2 ν = 0

H2 ν=1 -0.0044 -0.006 0.0069 -0.0157 D2 ν=1 0.0675

D2 ν = 1

0.8 E0 [eV]

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 34 of 51

0.7 0.6

exp. QD QCT MDEF

0.5 0.4

0

5

10

0

5

0

5

10

15 0

5

10

J

Figure 9: E0 (ν, J) parameters as a function of J for H2 and D2 reacting on Cu(211). Blue dots represent QCT results, red dots represent QD results, green dots represent MDEF results, and black dots represent experimental results 45 .

34

ACS Paragon Plus Environment

Page 35 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

3.D Classical molecular beam simulations One of the goals of this project was to carry out a molecular beam simulation using the QD method. Since surface atom motion and ehp excitations cannot be incorporated in QD calculations we have also performed molecular beam simulations using the BOMD, QCT and MDEF methods for D2 impinging on Cu(211) in order to quantify their effects on the reactivity measured in a molecular beam experiment. As discussed together with the comparison between our state resolved reaction probabilities and the associative desorption experiments of Kaufmann et al. 45 there are some effects on the reactivity from surface atom motion and ehp excitations though the effect falls within chemical accuracy. The molecular beam experiments we treat here were carried for a surface temperature of 120K 15,34 .

In figure 10 we compare BOMD calculations performed for a surface temperature of 120K (red) to QCT (black) and MDEF (green) calculations carried out on our six-dimensional PES. As an additional validation of the PES we have also calculated one energy point using the BOMD method with a rigid surface (blue). Each BOMD point is based on five hundred trajectories, each QCT and MDEF point on a hundred thousand trajectories. The molecular beam parameters were taken from refs 15,35 and can be found in table 2. From the excellent agreement in figure 10 between the black and blue data points at 80.1 kJ/mol it is clear that our PES was accurately fitted, as was previously demonstrated in figure S2 of ref 46 . There we showed that for the dynamically relevant region of the PES (VM AX < 2eV) the PES has a RMSE < 0.035 eV. Therefore results obtained from QD calculations performed on our PES should not be influenced much by any (small) lingering inaccuracies still present in the PES related to the fitting procedure. It can also be observed from figure 10 that the effect of surface motion is small, and well within the limits of chemical accuracy with respect to incidence energy. Due to the fact that H2 has a lower mass, we expect the effect of including surface motion during the dynamics will be even less pronounced for H2 than for D2 . We should also note here that when low surface temperature experiments are considered, as with 35

ACS Paragon Plus Environment

The Journal of Physical Chemistry

the 120K surface temperature here, it is known from the literature that the BOSS model works well for activated H2 dissociation on metals 26,33,34,36,60 .

It can also be seen from figure 10 that including the effect of ehp’s as a classical friction force shifts the reaction probability curve slightly to higher energies, and that the effect is rather small and linear with respect to the average translational energy. From the literature it is also known that including ehp excitations in the dynamics of H2 reacting on Cu(111) has only a marginal effect on the reaction probability 25,32,36,83 .

Due to the very small contribution of surface atom motion, and non-adiabatic effects incorporated in the MDEF calculations to the overall reaction probability, we pose that H2 impinging on Cu(211) is an excellent system to fully simulate a molecular beam experiment using quantum dynamics methods since large discrepancies between theory and experiment can reasonably be attributed to quantum effects during the dynamics, as the BOSS model should be quite accurate.

reaction probability

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

0.25

D2 TN QCT MDEF BOMD TS = 120K

= 2100K

BOMD rigid

0.2 0.15 0.1

65 70 75 80 average translational energy [kJ/mol]

Figure 10: Reaction probability as function of the average translational energy for D2 on Cu(211), with molecular beam parameters taken from table 2. BOMD results with a surface temperature of 120K are shown in red, MDEF results are shown in green, QCT results are shown in black. The blue point is an BOMD result for D2 on Cu(211) with a rigid surface.

36

ACS Paragon Plus Environment

Page 36 of 51

Page 37 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

3.E Quantum molecular beam simulations Figure 11 shows results of simulations for four sets of molecular beam experiments, with varying molecular beam conditions. The experiment of Rendulic and coworkers 14 has the broadest translational energy distributions. The molecular beam parameters are taken from (the supporting information of) refs. 15,34,35 Here, theoretical results obtained for the H2 + Cu(211) system are compared to theoretical results for the H2 + Cu(111) system, where for all theoretical results the SRP48 density functional was used. We only make a comparison to theoretical work since, to the best of our knowledge, there exists no published experimental molecular beam dissociative adsorption data for H2 reacting on Cu(211).

In order to make the best possible comparison between the QCT and QD results, both results are calculated from initial-state-resolved reaction probabilities for the same set of initial states. The molecular beam reaction probabilities predicted by QCT and QD calculations are in excellent agreement (figure 11). The excellent agreement holds for the very broad molecular beams of Rendulic and coworkers in figure 11a, as well as for the translationally narrow molecular beams of Auerbach and coworkers 15 shown in figures 11b-d. However, QCT predicts slightly higher reaction probabilities, especially for the lowest translational energies. The consistently higher QCT reaction probability can be attributed to zero-point energy (ZPE) leakage, which is not possible by design in the QD calculations wherein the ZPE is preserved.

The excellent agreement between the QCT and QD calculations implies that on the scale of a molecular beam experiment, in which a large number of rovibrational states are populated, quantum effects during the dynamics affect the reaction probability only in a very limited manner for reaction probabilities > 0.1%. The similarity between the QCT and QD calculations also holds over a wide range of molecular beam conditions, ranging from high to low incidence energies and from high to low nozzle temperatures. 37

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

From figure 11 it is also clear that for most incidence energies (> 22 kJ/mol) Cu(211) is predicted to be less reactive then Cu(111), as was reported previously for D2 + Cu(211) 46 . The lower reactivity of Cu(211) compared to Cu(111) cannot be explained by the d-band model 53,54 . In our previous paper we and others showed that the d-band model does make accurate predictions of the reactivity of different facets when similar reaction geometries are considered but that the breakdown of the predictive prowess of the d-band model is caused by the geometric effect of the lowest barrier to reaction for H2 dissociation on the low index Cu(111) surface not being on a top site.

Based on the results in figure 11 we can now say definitively that, on the scale of a molecular beam experiment, neglect of quantum effects during the dynamics cannot be invoked to explain the lower reactivity of Cu(211) than of Cu(111). This corroborates the theoretical results obtained in previous work 44,46 , where QCT calculations were performed for D2 and H2 , and S0 were measured for D2 + Cu(111) for Ei > 27 kJ/mol. More generally we can state that molecular beam sticking of H2 on cold Cu(211) is well described with quasiclassical dynamics, and this very probably also holds for H2 reacting on Cu(111) and Cu(100).

38

ACS Paragon Plus Environment

Page 38 of 51

Page 39 of 51

0.1

0.1

(b) pure H2

(a) pure H2 Rendulic et al.

0.01

reaction probability

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

0.001

0.01

25

30

0.001 30

35

35

40

45

50

0.1 (d) TN = 2000K

(c) TN = 1740K 0.01 0.01

QCT Cu(111) QCT Cu(211) QD Cu(211) 0.001

20

25

30

35

40

0.001

20

30

40

average translational energy [kJ/mol]

Figure 11: Comparison between four sets of molecular beam simulations for H2 + Cu(111) and Cu(211), using the SRP48 functional, and for normal incidence. Reactivity is shown as a function of average translational energy. The red dots correspond to QCT calculations for H2 + Cu(111). The green and blue dots correspond to, respectively, the QCT and QD calculations for H2 + Cu(211).

39

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

4. Conclusions In this work we have carried out a comprehensive study of the quantum reaction dynamics of H2 reacting on the Cu(211) surface. A large number of TDWP calculations have been performed for all important individual rovibrational states reasonably populated in a molecular beam experiment. Our main conclusion is that the reaction of H2 (D2 ) with Cu(211) is well described classically. This is especially true when simulating molecular beam experiments where one averages over a large number of rovibrational states and molecular beam energy distributions.

We have however found that the extent to which the reaction depends on the alignment of H2 is somewhat dependent on whether QD or the QCT method is used, requiring a careful validation of the dynamical model depending on the type of experiment that is being simulated. The QD method predicts stronger alignment effects on the reactivity than the QCT method for low lying rotational states.

A comparison to recent associative desorption experiments suggest and BOMD calculations appear to show that the effect of surface atom motion and ehp’s on the reactivity falls within chemical accuracy, even for the high surface temperature used in the associative desorption experiments. We saw no evidence in our fully-state-resolved data for the recently reported ’slow’ reaction channel, even though we carried out calculations over a translational energy range were this reported reactivity should be manifest. We speculate that the ’slow’ reaction channel is related to surface atom motion and that its modelling requires the description of this motion, which is why we did not see it here.

In contrast to the theoretical and experimental results for D2 reacting on Cu(111) and Cu(100), at low translational energy we observe a sharp downturn of the rotational quadrupole alignment parameters for vibrationally excited molecules. This downturn can be attributed 40

ACS Paragon Plus Environment

Page 40 of 51

Page 41 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

to a site specific reaction mechanism of inelastic rotational enhancement.

Acknowledgements This work was supported financially by the European Research Council though an ERC-2013 advanced grant (no. 338580) and with computer time granted by NWO-EC.

References (1) Wolcott, C. A.; Medford, A. J.; Studt, F.; Campbell, C. T. Degree of rate control approach to computational catalyst screening. J. Catal. 2015, 330, 197 – 207. (2) Sabbe, M. K.; Reyniers, M.-F.; Reuter, K. First-principles kinetic modeling in heterogeneous catalysis: an industrial perspective on best-practice, gaps and needs. Catal. Sci. Technol. 2012, 2, 2010–2024. (3) Ertl, G. Reactions at surfaces: from atoms to complexity (Nobel lecture). Angew. Chem. Int. Ed. 2008, 47, 3524–3535. (4) Waugh, K. Methanol synthesis. Catal. Today 1992, 15, 51–75. (5) Grabow, L.; Mavrikakis, M. Mechanism of methanol synthesis on Cu through CO2 and CO hydrogenation. Acs Catalysis 2011, 1, 365–384. (6) Behrens, M.; Studt, F.; Kasatkin, I.; K¨ uhl, S.; H¨avecker, M.; Abild-Pedersen, F.; Zander, S.; Girgsdies, F.; Kurr, P.; Kniep, B.-L. et al. The active site of methanol synthesis over Cu/ZnO/Al2 O3 industrial catalysts. Science 2012, 336, 893–897. (7) Zambelli, T.; Wintterlin, J.; Trost, J.; Ertl, G. Identification of the” active sites” of a surface-catalyzed reaction. Science 1996, 273, 1688–1690.

41

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(8) Somorjai, G.; Joyner, R.; Lang, B. The reactivity of low index [(111) and (100)] and stepped platinum single crystal surfaces. Proc. Royal Soc. Lond. A. 1972, 331, 335–346. (9) Kroes, G.-J. Toward a database of chemically accurate barrier heights for reactions of molecules with metal surfaces. J. Phys. Chem lett. 2015, 6, 4106–4114. (10) Polanyi, J. C. Some concepts in reaction dynamics. Science 1987, 236, 680–690. (11) Polanyi, J. C. Some concepts in reaction dynamics (Nobel lecture). Angew. Chem. 1987, 26, 952–971. ˇ ˇ Hammer, B. H2 dissociation at defected Cu: preference for reaction at (12) Sljivanˇ canin, Z.; vacancy and kink sites. Phys. Rev. B 2002, 65, 085414. (13) Anger, G.; Winkler, A.; Rendulic, K. Adsorption and desorption kinetics in the systems H2 /Cu(111), H2 /Cu(110) and H2 /Cu(100). Surf. Sci. 1989, 220, 1–17. (14) Berger, H.; Leisch, M.; Winkler, A.; Rendulic, K. A search for vibrational contributions to the activated adsorption of H2 on copper. Chem. Phys. Lett. 1990, 175, 425–428. (15) Michelsen, H.; Rettner, C.; Auerbach, D.; Zare, R. Effect of rotation on the translational and vibrational energy dependence of the dissociative adsorption of D2 on Cu(111). J. Chem. Phys. 1993, 98, 8294–8307. (16) Hou, H.; Gulding, S.; Rettner, C.; Wodtke, A.; Auerbach, D. The stereodynamics of a gas-surface reaction. Science 1997, 277, 80–82. (17) Comsa, G.; David, R. The purely fast distribution of H2 and D2 molecules desorbing from Cu(100) and Cu(111) surfaces. Surf. Sci. 1982, 117, 77–84. (18) Rettner, C.; Michelsen, H.; Auerbach, D. Quantum-state-specific dynamics of the dissociative adsorption and associative desorption of H2 at a Cu(111) surface. J. Chem. Phys. 1995, 102, 4625–4641. 42

ACS Paragon Plus Environment

Page 42 of 51

Page 43 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

(19) Rettner, C.; Michelsen, H.; Auerbach, D. Determination of quantum-state-specific gassurface energy transfer and adsorption probabilities as a function of kinetic energy. Chem. Phys. 1993, 175, 157–169. (20) Gostein, M.; Parhikhteh, H.; Sitz, G. Survival probability of H2 (v= 1, J= 1) scattered from Cu(110). Phys. Rev. Lett. 1995, 75, 342. (21) Hodgson, A.; Samson, P.; Wight, A.; Cottrell, C. Rotational excitation and vibrational relaxation of H2 (υ= 1, J= 0) scattered from Cu(111). Phys. Rev. Lett. 1997, 78, 963. (22) Watts, E.; Sitz, G. O. State-to-state scattering in a reactive system: H2 (v= 1, j= 1) from Cu(100). J. Chem. Phys. 2001, 114, 4171–4179. (23) Rettner, C.; Auerbach, D.; Michelsen, H. Dynamical studies of the interaction of D2 with a Cu(111) surface. J. Vac. Sci. Technol. A 1992, 10, 2282–2286. (24) Michelsen, H.; Rettner, C.; Auerbach, D. On the influence of surface temperature on adsorption and desorption in the D2 /Cu(111) system. Surf. Sci. 1992, 272, 65–72. (25) Spiering, P.; Meyer, J. Testing electronic friction models: vibrational de-excitation in scattering of H2 and D2 from Cu(111). J. Phys. Chem. Lett. 2018, 9, 1803–1808. (26) Spiering, P.; Wijzenbroek, M.; Somers, M. An improved static corrugation model. J. Chem. Phys. 2018, 149, 234702. (27) Kroes, G.-J.; Wiesenekker, G.; Baerends, E.; Mowrey, R. Competition between vibrational excitation and dissociation in collisions of H2 with Cu(100). Phys. Rev. B 1996, 53, 10397. (28) Kroes, G.-J.; Juaristi, J.; Alducin, M. Vibrational excitation of H2 scattering from Cu(111): effects of surface temperature and of allowing energy exchange with the surface. J. Phys. Chem. C 2017, 121, 13617–13633.

43

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(29) Salin, A. Theoretical study of hydrogen dissociative adsorption on the Cu(110) surface. J. Chem. Phys. 2006, 124, 104704. (30) Sementa, L.; Wijzenbroek, M.; Van Kolck, B.; Somers, M.; Al-Halabi, A.; Busnengo, H. F.; Olsen, R.; Kroes, G.-J.; Rutkowski, M.; Thewes, C. et al. Reactive scattering of H2 from Cu(100): comparison of dynamics calculations based on the specific reaction parameter approach to density functional theory with experiment. J. Chem. Phys. 2013, 138, 044708. (31) Somers, M.; McCormack, D.; Kroes, G.-J.; Olsen, R.; Baerends, E.; Mowrey, R. Signatures of site-specific reaction of H2 on Cu(100). J. Chem. Phys. 2002, 117, 6673–6687. (32) Nattino, F.; D´ıaz, C.; Jackson, B.; Kroes, G.-J. Effect of surface motion on the rotational quadrupole alignment parameter of D2 reacting on Cu(111). Phys. Rev. Lett. 2012, 108, 236104. (33) Kroes, G.-J.; D´ıaz, C. Quantum and classical dynamics of reactive scattering of H2 from metal surfaces. Chem. Soc. Rev. 2016, 45, 3658–3700. (34) D´ıaz, C.; Olsen, R. A.; Auerbach, D. J.; Kroes, G.-J. Six-dimensional dynamics study of reactive and non reactive scattering of H2 from Cu(111) using a chemically accurate potential energy surface. Phys. Chem. Chem. Phys. 2010, 12, 6499–519. (35) D´ıaz, C.; Pijper, E.; Olsen, R. A.; Busnengo, H. F.; Auerbach, D. J.; Kroes, G.-J. Chemically accurate simulation of a prototypical surface reaction: H2 dissociation on Cu(111). Science 2009, 326, 832–834. (36) Wijzenbroek, M.; Somers, M. F. Static surface temperature effects on the dissociation of H2 and D2 on Cu(111). J. Chem. Phys. 2012, 137, 054703. (37) McCormack, D. A.; Kroes, G.-J.; Olsen, R. A.; Groeneveld, J. A.; van Stralen, J. N.; Baerends, E. J.; Mowrey, R. C. Quantum dynamics of the dissociation of H2 on Cu(100): 44

ACS Paragon Plus Environment

Page 44 of 51

Page 45 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

dependence of the site-reactivity on initial rovibrational state. Faraday Discuss. 2000, 117, 109–132. (38) McCormack, D. A.; Kroes, G.-J.; Olsen, R. A.; Groeneveld, J. A.; van Stralen, J. N.; Baerends, E. J.; Mowrey, R. C. Molecular knife throwing: aiming for dissociation at specific surface sites through state-selection. Chem. Phys. Lett. 2000, 328, 317–324. (39) Chen, J.; Zhou, X.; Jiang, B. Eley Rideal recombination of hydrogen atoms on Cu(111): quantitative role of electronic excitation in cross sections and product distributions. J. Chem. Phys. 2019, 150, 061101. (40) Darling, G.; Holloway, S. Rotational motion and the dissociation of H2 on Cu(111). J. Chem. Phys. 1994, 101, 3268–3281. (41) Nienhaus, H.; Bergh, H.; Gergen, B.; Majumdar, A.; Weinberg, W.; McFarland, E. Electron-hole pair creation at Ag and Cu surfaces by adsorption of atomic hydrogen and deuterium. Phys. Rev. Lett. 1999, 82, 446. (42) Sakong, S.; Groß, A. Dissociative adsorption of hydrogen on strained Cu surfaces. Surf. Sci. 2003, 525, 107–118. (43) Kroes, G.-J.; Pijper, E.; Salin, A. Dissociative chemisorption of H2 on the Cu(110) surface: A quantum and quasiclassical dynamical study. J. Chem. Phys. 2007, 127, 164722. (44) Cao, K.; F¨ uchsel, G.; Kleyn, A. W.; Juurlink, L. B. Hydrogen adsorption and desorption from Cu(111) and Cu(211). Phys. Chem. Chem. Phys. 2018, 20, 22477–22488. (45) Kaufmann, S.; Shuai, Q.; Auerbach, D. J.; Schwarzer, D.; Wodtke, A. M. Associative desorption of hydrogen isotopologues from copper surfaces: characterization of two reaction mechanisms. J. Chem. Phys. 2018, 148, 194703.

45

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(46) F¨ uchsel, G.; Cao, K.; Er, S.; Smeets, E. W. F.; Kleyn, A. W.; Juurlink, L. B. F.; Kroes, G.-J. Anomalous dependence of the reactivity on the presence of steps: dissociation of D2 on Cu(211). J. Phys. Chem. Lett. 2018, 9, 170–175. (47) Ghassemi, E. N.; Smeets, E. W. F.; Somers, M. F.; Kroes, G.-J.; Groot, I. M.; Juurlink, L. B.; F¨ uchsel, G. Transferability of the specific reaction parameter density functional for H2 + Pt(111) to H2 + Pt(211). J. Phys. Chem. C 2019, 123, 2973–2986. (48) Olsen, R.; McCormack, D.; Baerends, E. How molecular trapping enhances the reactivity of rough surfaces. Surf. Sci. 2004, 571, L325–L330. (49) McCormack, D. A.; Olsen, R. A.; Baerends, E. J. Mechanisms of H2 dissociative adsorption on the Pt(211) stepped surface. J. Chem. Phys. 2005, 122, 194708. (50) Luppi, M.; McCormack, D. A.; Olsen, R. A.; Baerends, E. J. Rotational effects in the dissociative adsorption of H2 on the Pt(211) stepped surface. J. Chem. Phys. 2005, 123, 164702. (51) Olsen, R.; McCormack, D.; Luppi, M.; Baerends, E. Six-dimensional quantum dynamics of H2 dissociative adsorption on the Pt(211) stepped surface. J. Chem. Phys. 2008, 128, 194715. (52) Huang, X.; Yan, X.; Xiao, Y. Effects of vacancy and step on dissociative dynamics of H2 on Pd(111) surfaces. Chem. Phys. Lett. 2012, 531, 143–148. (53) Hammer, B.; Nørskov, J. Electronic factors determining the reactivity of metal surfaces. Surf. Sci. 1995, 343, 211 – 220. (54) Hammer, B.; Nørskov, J. Why gold is the noblest of all the metals. Nature 1995, 376, 238. (55) Mavrikakis, M.; Hammer, B.; Nørskov, J. K. Effect of strain on the reactivity of metal surfaces. Phys. Rev. Lett. 1998, 81, 2819. 46

ACS Paragon Plus Environment

Page 46 of 51

Page 47 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

(56) Pijper, E.; Kroes, G.-J.; Olsen, R. A.; Baerends, E. J. Reactive and diffractive scattering of H2 from Pt(111) studied using a six-dimensional wave packet method. J. Chem. Phys. 2002, 117, 5885–5898. (57) Kroes, G.-J.; Somers, M. F. Six-dimensional dynamics of dissociative chemisorption of H2 on metal surface. J. Theor. Comput. Chem. 2005, 04, 493–581. (58) Wijzenbroek, M.; Klein, D. M.; Smits, B.; Somers, M. F.; Kroes, G.-J. Performance of a non-local van der Waals density functional on the dissociation of H2 on metal surfaces. J. Phys. Chem. A 2015, 119, 12146–12158. (59) Busnengo, H.; Salin, A.; Dong, W. Representation of the 6D potential energy surface for a diatomic molecule near a solid surface. J. Chem. Phys. 2000, 112, 7641–7651. (60) Wei, C. Y.; Lewis, S. P.; Mele, E. J.; Rappe, A. M. Structure and vibrations of the vicinal copper (211) surface. Phys. Rev. B 1998, 57, 10062–10068. (61) Hammer, B.; Hansen, L. B.; Nørskov, J. K. Improved adsorption energetics within density-functional theory using revised Perdew-Burke-Ernzerhof functionals. Phys. Rev. B 1999, 59, 7413–7421. (62) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized gradient approximation made simple. Phys. Rev. Lett. 1996, 77, 3865–3868. (63) Sementa, L.; Wijzenbroek, M.; van Kolck, B. J.; Somers, M. F.; Al-Halabi, A.; Busnengo, H. F.; Olsen, R. A.; Kroes, G.-J.; Rutkowski, M.; Thewes, C. et al. Reactive scattering of H2 from Cu(100): comparison of dynamics calculations based on the specific reaction parameter approach to density functional theory with experiment. J. Chem. Phys. 2013, 138, 044708. (64) Kresse, G.; Hafner, J. Ab initio molecular-dynamics simulation of the liquid-metal–

47

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

amorphous-semiconductor transition in germanium. Phys. Rev. B 1994, 49, 14251– 14269. (65) Kresse, G.; Hafner, J. Ab initio molecular dynamics for liquid metals. Phys. Rev. B 1993, 47, 558–561. (66) Kresse, G.; Furthm¨ uller, J. Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set. Phys. Rev. B 1996, 54, 11169–11186. (67) Kresse, G.; Furthm¨ uller, J. Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set. Comput. Mater. Sci. 1996, 6, 15–50. (68) Blanco-Rey, M.; Juaristi, J. I.; D´ıez Mui˜ no, R.; Busnengo, H. F.; Kroes, G.-J.; Alducin, M. Electronic friction dominates hydrogen hot-atom relaxation on Pd(100). Phys. Rev. Lett. 2014, 112, 103203. (69) Novko, D.; Blanco-Rey, M.; Juaristi, J. I.; Alducin, M. Ab initio molecular dynamics with simultaneous electron and phonon excitations: Application to the relaxation of hot atoms and molecules on metal surfaces. Phys. Rev. B 2015, 92, 201411. (70) Michelsen, H. A.; Rettner, C. T.; Auerbach, D. J.; Zare, R. N. Effect of rotation on the translational and vibrational energy dependence of the dissociative adsorption of D2 on Cu(111). J. Chem. Phys. 1993, 98, 8294–8307. (71) F¨ uchsel, G.; del Cueto, M.; D´ıaz, C.; Kroes, G.-J. Enigmatic HCl + Au(111) reaction: a puzzle for theory and experiment. J. Phys. Chem. C 2016, 120, 25760–25779. (72) Raff, L. M.; Karplus, M. Theoretical investigations of reactive collisions in molecular beams: K+CH3 I and related systems. J. Chem. Phys. 1966, 44, 1212–1229. (73) Lemons, D. S.; Gythiel, A. On the Theory of Brownian Motion. Am. J. Phys. 1997, 65, 1079–1081.

48

ACS Paragon Plus Environment

Page 48 of 51

Page 49 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

(74) Ermak, D. L.; Buckholz, H. Numerical integration of the Langevin equation: Monte Carlo simulation. J. Comput. Phys. 1980, 35, 169–182. (75) F¨ uchsel, G.; Klamroth, T.; Monturet, S.; Saalfrank, P. Dissipative dynamics within the electronic friction approach: the femtosecond laser desorption of H2 /D2 from Ru(0001). Phys. Chem. Chem. Phys. 2011, 13, 8659–8670. (76) Juaristi, J.; Alducin, M.; Mui˜ no, R. D.; Busnengo, H. F.; Salin, A. Role of electron-hole pair excitations in the dissociative adsorption of diatomic molecules on metal surfaces. Phys. Rev. Lett. 2008, 100, 116102. (77) Vibok, A.; Balint-Kurti, G. Parametrization of complex absorbing potentials for timedependent quantum dynamics. J. Phys. Chem. 1992, 96, 8712–8719. (78) Balint-Kurti, G. G.; Dixon, R. N.; Marston, C. C. Grid methods for solving the Schr¨odinger equation and time dependent quantum dynamics of molecular photofragmentation and reactive scattering processes. Int. Rev. Phys. Chem. 1992, 11, 317–344. (79) Michelsen, H. A.; Auerbach, D. J. A critical examination of data on the dissociative adsorption and associative desorption of hydrogen at copper surfaces. J. Chem. Phys. 1991, 94, 7502–7520. (80) Zare, R. N.; Harter, W. G. Angular momentum: understanding spatial aspects in chemistry and physics. Phys. Today 1989, 42, 68. (81) Shakouri, K.; Behler, J.; Meyer, J.; Kroes, G.-J. Accurate neural network description of surface phonons in reactive gas-surface dynamics: N2 + Ru(0001). J. Phys. Chem. Lett. 2017, 8, 2131–2136. (82) D´ıaz, C.; Perrier, A.; Kroes, G.-J. Associative desorption of N2 from Ru(0001): a computational study. Chem. Phys. Lett. 2007, 434, 231–236.

49

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(83) Muzas, A. S.; Juaristi, J. I.; Alducin, M.; D´ıez Mui˜ no, R.; Kroes, G.-J.; D´ıaz, C. Vibrational deexcitation and rotational excitation of H2 and D2 scattered from Cu(111): adiabatic versus non-adiabatic dynamics. J. Chem. Phys. 2012, 137, 064707.

50

ACS Paragon Plus Environment

Page 50 of 51

Page 51 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Graphical TOC Entry

51

ACS Paragon Plus Environment